Aditya Iyer Email & Phone Number

California, United States

Berkeley, California, United States

Berkeley

Developing an open source speech to text model from scratch.

Westborough, Massachusetts, United States

• Engineered an advanced search engine using Sentence Transformers to generate vector embeddings and compute cosine similarity for over 100,000 data points.
• Created RESTful API’s using FastAPI that enabled precise retrieval of CPT, ICD, and SNOMED codes from natural language input, with advanced filters for age and gender.
• Developed an algorithm to classify medical documents, achieving 100% test accuracy. Utilized TFIDF vectorization and a retrainable Stochastic Gradient Descent (SGD) classifier, ensuring seamless integration with.
• Leveraged Pytesseract and OpenCV to extract text and tables from medical documents, and integrated these with OpenAI’s API to enable sophisticated document question answering.
• Developed a predictive model trained using the Random Forest Classifier on over 1million data points to successfully predict whether a health insurance claim will be denied with 98% accuracy.
• Created an internal tool that can extract business requirements from large documents, saving over 2 hours of manual work.

• Expertly processed neuroimaging MRI data through advanced techniques including midpoint slicing along axes, reshaping axes for uniformity, and applying min-max scaling to optimize image generation.
• Pioneered the use of Zero Shot Image Classification to accurately predict regions of brain activity, demonstrating an innovative approach to neuroimaging analysis.
• Conducted comprehensive performance testing of the Independent Component Analysis model across CPU and GPU environments, ensuring robustness and efficiency in computational neuroscience studies.

• Collaborated with a startup to execute comprehensive data pre-processing and exploratory data analysis (EDA) on socialmedia datasets, employing advanced NLP techniques.
• Conducted sentiment analysis using VADER on Twitter data to develop a sentiment analysis and political stance detectionmodel with a dataset of over 10,000 data points.
• Led the consolidation of diverse datasets with over 25,000 data points, creating a prompt-string construction methodologybased on a research paper, optimized for GPT-4 and Llama-2 models, enhancing generative AI.

• Leveraged advanced machine learning techniques to pioneer the development of an innovative BERT model for Ribosomal RNA sequencing prediction, collaborating with Nobel Laureate Jennifer Doudna’s Team and Professor.
• Developed a tokenizer to split data into single, double, and triple NT sequences and compare the Jensen-Shannon Divergence (JSD).
• Trained and evaluated the model on over 1000000 23s-rRNA data points using PyTorch, scikit-learn, and regex, achieving an exceptionally low Jensen-Shannon Divergence (JSD) score of 0.005, which demonstrates the model’s.

Served as a Lab Academic Intern for Berkeley's Data Structures and Algorithms Course. Leading lab sessions of 30+ students from various backgrounds, reinforcing CS data structure and algorithm concepts such as trees, linked lists, search, sort and other Java data structures. Assist students to debug their code on lab and homework assignments.

Served as a Lab Academic Intern for one of the largest classes at Berkeley with over 2000+ students. Leading lab sessions of 30+ students from various backgrounds, reinforcing CS fundamental concepts such as recursion, trees, object-oriented programming, linked lists, and other python data structures. Assist students to debug their code on lab and homework.

Working as a Lab Academic Intern for the introductory CS class at Berkeley. Helping students debug their code and understand fundamental coding concepts and data structures.

Berkeley, California, United States

Conducted weekly one-hour instructional sessions for a cohort of five to six students currently enrolled in the course CS61A. These sessions involved the curation and delivery of concise lectures designed to complement the material presented by the course professor. Additionally, I facilitated discussions surrounding meticulously crafted supplemental.

Berkeley, California, United States

Led weekly 1-hour sessions with 5-6 students currently enrolled in CS61A. Curated and presented mini-lectures to cover corresponding material taught by the Professor. Go over specially curated worksheets for CSM and answer any questions that students might have!

I spent my time at MetaCoders helping them to educate students in coding through online workshops, assisting in administrative tasks, and editing/creating a computer science curriculum.

Bachelor'S Degree, Data Science (Applied Math Emphasis), Statistics (Physics Concentration)

Activities and Societies: ULAB (Physics and Astronomy Department), Quantum Computing at Berkeley, Computer Science Mentors @ Berkeley.

High School Diploma, Ib

Activities and Societies: Model UN Club, Karate, Self Defense Club, Pegasus fundraising team, Community Kitchen Team lead, Panel.

High School Diploma, Physics, 95%

Activities and Societies: Football team captain, house vice-captain- student council, fundraising, and organizing team, Competitive Model.